Say "skipped" in scan's warnings for skipped paths (closes #170)

A warning for a path the summary counts as skipped now reads
`scan: skipped PATH (REASON)`, naming the path once. This covers
unreadable directories, failed lstats, failed reads and dropped
operands. Each path of a hard-linked file that cannot be read gets its
own line, since each is counted. Lines for paths left out on purpose
(`--exclude`, default excludes, filesystems that are not persistent)
now read `scan: left out PATH (...)`, so a search for `scan: skipped`
finds only the counted paths.

Model: opus-5-5

README.md

@@ -29,7 +29,6 @@ first look at a new machine with one command.
 
 ```sh
 make build
-export SFDUPES_DATABASE="$HOME/.local/share/sfdupes/db.sqlite"
 ./sfdupes scan /srv
 ./sfdupes report > dupes.tsv
 ./sfdupes trees > dupetrees.tsv
@@ -47,8 +46,10 @@ Both cover the whole database, or only the paths given after them, as in
 `sfdupes report /srv/photos`. `info` prints the database's path, size and record
 count.
 
-The database is `/var/lib/sfdupes/db.sqlite` unless `SFDUPES_DATABASE` names
+Root keeps the database in `/var/lib/sfdupes/db.sqlite`. Any other user keeps it
-another file. The first `scan` creates it.
+in `sfdupes/db.sqlite` in the XDG state directory, `~/.local/state` by default
+on Linux. `SFDUPES_DATABASE` names another file (see "Location"). The first
+`scan` creates the database.
 
 `report`, `trees` and `info` write their data to stdout. Progress, warnings and
 errors go to stderr. stdout and stderr are a program's two output streams. The
@@ -135,8 +136,8 @@ point of each disk at or below `/srv` with its own `mountpoint -q`, joined with
     - one line naming the database;
     - progress lines, as described under "Progress";
     - a warning for each path it skips;
-    - a line for each system or temporary directory the walk leaves out (see
+    - a line for each system, temporary or cache directory the walk leaves out
-      "Rules for the walk");
+      (see "Rules for the walk");
     - a line for each path left out because its filesystem is one the kernel,
       the core of the operating system, makes up or keeps in memory, such as
       `tmpfs`;
@@ -155,8 +156,9 @@ point of each disk at or below `/srv` with its own `mountpoint -q`, joined with
 - `report`, `trees` and `info` need only read access to the database. The umask
   is the setting that removes permissions from the files a process creates.
   Under the usual umask of `022`, the first scan creates the database readable
-  by every user. An unprivileged user can then run those three commands against
+  by every user. Another user can then run those three commands against root's
-  root's database.
+  database, with `SFDUPES_DATABASE=/var/lib/sfdupes/db.sqlite` before the
+  command.
 
 ### Reading the reports
 
@@ -295,9 +297,9 @@ The reference below is also the program's specification.
 ### Settings
 
 - `SFDUPES_DATABASE` names the database file. When it is unset or empty, the
-  database is `/var/lib/sfdupes/db.sqlite`. There is no command-line flag for
+  database is at its default path (see "Location"). There is no command-line
-  it. The path names the file exactly, whatever characters it holds. A relative
+  flag for it. The path names the file exactly, whatever characters it holds. A
-  path is relative to the working directory.
+  relative path is relative to the working directory.
 - `--workers N` (`scan`, `dwim`) sets how many directories `scan` reads at once
   while it lists the trees. It also sets how many files `scan` reads at once
   while it hashes them. The default is the number of CPUs. `N` must be at
@@ -307,8 +309,9 @@ The reference below is also the program's specification.
   of directories on other filesystems, as the same flag does for `du` and
   `rsync`. By default the walk also enters directories on other filesystems.
 - `--no-default-excludes` (`scan`, `dwim`) lets the walk enter the default
-  excludes. Those are the system and temporary directories the walk otherwise
+  excludes. Those are the system, temporary and cache directories the walk
-  leaves out, such as `/var/tmp` on Linux (see "Rules for the walk").
+  otherwise leaves out, such as `/var/tmp` and `/var/cache` on Linux (see "Rules
+  for the walk").
 - `--exclude PATH` (`scan`, `dwim`) keeps the walk out of `PATH` and everything
   below it. `PATH` may be a directory or a file. The flag may be given more than
   once. The walk knows `PATH` by its device ID and inode number, as it knows the
@@ -560,7 +563,7 @@ walk practical on large or busy storage pools.
     `scan` prints a line on stderr naming the path and its filesystem type:
 
     ```
-    scan: skipped /dev/shm (tmpfs, not a persistent filesystem)
+    scan: left out /dev/shm (tmpfs, not a persistent filesystem)
     ```
 
     The path is not counted as skipped. The records below it are kept, as
@@ -570,7 +573,7 @@ walk practical on large or busy storage pools.
   For each one it reaches, `scan` prints a line on stderr naming it:
 
     ```
-    scan: skipped /srv/cache (--exclude)
+    scan: left out /srv/cache (--exclude)
     ```
 
     The path is not counted as skipped. The scan keeps the records at and below
@@ -578,12 +581,17 @@ walk practical on large or busy storage pools.
     `--no-default-excludes` does not change this rule.
 
 - Below an operand, the walk does not enter a default exclude. On Linux, the
-  default excludes hold temporary files and the kernel's view of the running
+  default excludes hold the kernel's view of the running system, temporary
-  system. They depend on the system, and `sfdupes scan --help` lists those of
+  files, and caches. A cache here is data that programs and package managers
-  the system it runs on. On Linux they are:
+  download or build again on their own. The default excludes depend on the
+  system, and `sfdupes scan --help` lists those of the system it runs on. In the
+  lists, `*` stands for any one directory name, so `/home/*/.cache` matches
+  `/home/alice/.cache`. On Linux they are:
 
     ```
-    /proc, /sys, /dev, /run, /tmp, /var/tmp
+    /proc, /sys, /dev, /run, /tmp, /var/tmp, /var/cache
+    /root/.cache, /root/.npm, /root/.cargo/registry, /root/go/pkg/mod
+    /home/*/.cache, /home/*/.npm, /home/*/.cargo/registry, /home/*/go/pkg/mod
     ```
 
     On macOS they are:
@@ -591,13 +599,21 @@ walk practical on large or busy storage pools.
     ```
     /dev, /private/tmp, /private/var/tmp, /private/var/vm
     /System/Volumes/VM, /System/Volumes/Data
+    /private/var/folders, /Library/Caches
+    /private/var/root/Library/Caches, /private/var/root/.cache
+    /private/var/root/.npm, /private/var/root/.cargo/registry
+    /private/var/root/go/pkg/mod
+    /Users/*/Library/Caches, /Users/*/.cache, /Users/*/.npm
+    /Users/*/.cargo/registry, /Users/*/go/pkg/mod
     ```
 
-    A walk of `/` on macOS reaches everything in `/System/Volumes/Data` already,
+    `/private/var/folders` holds each user's temporary files and caches on
-    through firmlinks. Firmlinks are the links macOS uses to show the data
+    macOS. Root's home on macOS is `/var/root`, which the walk reaches as
-    volume inside the system volume. Walking `/System/Volumes/Data` as well
+    `/private/var/root`. A walk of `/` on macOS reaches everything in
-    would record each of those files twice. Any other system has no default
+    `/System/Volumes/Data` already, through firmlinks. Firmlinks are the links
-    excludes.
+    macOS uses to show the data volume inside the system volume. Walking
+    `/System/Volumes/Data` as well would record each of those files twice. Any
+    other system has no default excludes.
 
     An operand at or inside a default exclude is walked, so
     `sfdupes scan /tmp/x` works when `/tmp/x` is on a persistent filesystem.
@@ -606,7 +622,7 @@ walk practical on large or busy storage pools.
     not enter, `scan` prints a line on stderr naming it:
 
     ```
-    scan: skipped /var/tmp (default exclude)
+    scan: left out /var/tmp (default exclude)
     ```
 
     The directory is not counted as skipped. `scan` prints the filesystem line
@@ -666,6 +682,18 @@ scan: 123400 files seen (1200 added, 34 updated, 56 removed, 122166 unchanged),
 
 `removed` counts deleted records, which are not part of the files seen.
 
+The warning for a skipped path names the path once and gives the reason:
+
+```
+scan: skipped /srv/a/closed (permission denied)
+scan: skipped /srv/a/pipe (a FIFO)
+```
+
+The summary counts each path named in such a warning. A line for a path left out
+on purpose starts with `scan: left out` instead, and is not counted. A search of
+the log for `scan: skipped` therefore finds the paths the summary counts. When a
+file with several hard links cannot be read, each of its paths gets a warning.
+
 #### Stopping a scan
 
 `scan` stops cleanly on `SIGINT`, the signal Ctrl-C sends, or `SIGTERM`, the
@@ -977,7 +1005,22 @@ All five subcommands work on one SQLite database file.
 #### Location
 
 The database is the file named by `SFDUPES_DATABASE`, when that is set and not
-empty, and otherwise `/var/lib/sfdupes/db.sqlite`.
+empty. Otherwise it is at its default path, which depends on the user:
+
+- For root, it is `/var/lib/sfdupes/db.sqlite`.
+- For any other user, it is `sfdupes/db.sqlite` in the XDG state directory. That
+  directory is `$XDG_STATE_HOME` when that is set and not empty. Otherwise it is
+  `~/.local/state` on Linux and `~/Library/Application Support` on macOS. An
+  `XDG_STATE_HOME` that is not an absolute path is a fatal error, exit 1, for
+  every subcommand. The error names `XDG_STATE_HOME`.
+
+`scan` creates each missing directory on the way to the database with mode 755.
+At the default path of a user other than root, it uses mode 700 instead, as the
+XDG Base Directory specification asks. Other users cannot enter a directory with
+mode 700, so they cannot read a database in it.
+
+All five subcommands use the same default, so a user's `report` reads the
+database that user's `scan` wrote.
 
 Each subcommand prints one line on stderr naming the database it uses, before
 anything else and whether or not stderr is a terminal. The line gives the path
@@ -1017,9 +1060,9 @@ Only one `scan` at a time can hold the lock.
 - For its whole run, `scan` holds a lock on a lock file beside the database. The
   lock is taken with the `flock(2)` system call, a request a program makes to
   the kernel, and only one process can hold it. The lock file's name is the
-  database's path, with symlinks resolved, plus `.lock`. By default that is
+  database's path, with symlinks resolved, plus `.lock`. For root, by default,
-  `/var/lib/sfdupes/db.sqlite.lock`. A path through a symlink names the same
+  that is `/var/lib/sfdupes/db.sqlite.lock`. A path through a symlink names the
-  lock file.
+  same lock file.
 - `scan` takes the lock before it walks the filesystem or opens the database.
   Just before that, it creates the database file, empty, if it is missing. So
   the lock file is named the same way before the first scan as after. A symlink

cmd/sfdupes/main.go

@@ -5,8 +5,10 @@
 // KiB, and of the content. The content hash covers the whole file under
 // 100 MiB, and 1 MiB at each gigabyte at 100 MiB or more, so a large
 // file is never read in full. scan maintains a persistent SQLite
-// database of file signatures (SFDUPES_DATABASE, default
+// database of file signatures that the other subcommands read. The
-// /var/lib/sfdupes/db.sqlite) that the other subcommands read.
+// database is SFDUPES_DATABASE, or by default /var/lib/sfdupes/db.sqlite
+// for root and sfdupes/db.sqlite in the XDG state directory for any
+// other user.
 //
 // Usage:
 //

go.mod

@@ -3,6 +3,7 @@ module sneak.berlin/go/sfdupes
 go 1.25.7
 
 require (
+	github.com/adrg/xdg v0.5.3
 	github.com/rivo/uniseg v0.4.7
 	github.com/schollz/progressbar/v3 v3.19.1
 	github.com/spf13/cobra v1.10.2

go.sum

@@ -1,3 +1,5 @@
+github.com/adrg/xdg v0.5.3 h1:xRnxJXne7+oWDatRhR1JLnvuccuIeCoBu2rtuLqQB78=
+github.com/adrg/xdg v0.5.3/go.mod h1:nlTsY+NNiCBGCK2tpm09vRqfVzrc2fLmXGpBLF0zlTQ=
 github.com/chengxilo/virtualterm v1.0.4 h1:Z6IpERbRVlfB8WkOmtbHiDbBANU7cimRIof7mk9/PwM=
 github.com/chengxilo/virtualterm v1.0.4/go.mod h1:DyxxBZz/x1iqJjFxTFcr6/x+jSpqN0iwWCOK1q10rlY=
 github.com/cpuguy83/go-md2man/v2 v2.0.6/go.mod h1:oOW0eioCTA6cOiMLiUPZOpcVxMig6NIQQ7OS05n1F4g=

internal/sfdupes/cli.go

@@ -34,10 +34,14 @@ const (
 
 // databaseHelp is the paragraph of the root and scan help that names
 // the database (README "Location").
-const databaseHelp = "scan and dwim store what they find in a database " +
+const databaseHelp = `scan and dwim store what they find in a database file. report,
-	"file. report, trees\nand info read it. The file is $" + databaseEnv +
+trees and info read it. The file is $SFDUPES_DATABASE when that is set
-	" when that is set and\nnot empty, and otherwise " +
+and not empty. Otherwise root uses /var/lib/sfdupes/db.sqlite. Any other
-	defaultDatabasePath + ", where creating it\nneeds root.\n"
+user uses sfdupes/db.sqlite in the XDG state directory: $XDG_STATE_HOME,
+or ~/.local/state when that is unset or empty
+(~/Library/Application Support on macOS). An XDG_STATE_HOME that is
+not an absolute path is an error.
+`
 
 // escapesHelp is the paragraph of the report and trees help on how
 // paths are escaped (README "Report output format").

internal/sfdupes/cli_test.go

@@ -15,6 +15,7 @@ import (
 	"strings"
 	"testing"
 
+	"github.com/adrg/xdg"
 	"golang.org/x/sys/unix"
 )
 
@@ -559,13 +560,17 @@ func TestRunNamesDefaultDatabase(t *testing.T) {
 	// line.
 	t.Setenv(databaseEnv, "")
 
-	// Whether scan calls the database new depends on that machine too.
+	dbPath, err := defaultDatabasePath(os.Geteuid())
-	scanWant := "scan: database " + defaultDatabasePath + " (default)\n"
+	if err != nil {
+		t.Fatal(err)
+	}
 
-	_, err := os.Stat(defaultDatabasePath)
+	// Whether scan calls the database new depends on that machine too.
+	scanWant := "scan: database " + dbPath + " (default)\n"
+
+	_, err = os.Stat(dbPath)
 	if errors.Is(err, fs.ErrNotExist) {
-		scanWant = "scan: new database " + defaultDatabasePath +
+		scanWant = "scan: new database " + dbPath + " (default)\n"
-			" (default)\n"
 	}
 
 	missing := filepath.Join(t.TempDir(), "nope")
@@ -585,21 +590,21 @@ func TestRunNamesDefaultDatabase(t *testing.T) {
 			run: func() error {
 				return runReport(cancelledContext(t), nil, io.Discard)
 			},
-			want: "report: database " + defaultDatabasePath + " (default)\n" +
+			want: "report: database " + dbPath + " (default)\n" +
 				reportOpenLine + "\n" + reportOpenLine + "\n",
 		},
 		cmdTrees: {
 			run: func() error {
 				return runTrees(cancelledContext(t), nil, io.Discard)
 			},
-			want: "trees: database " + defaultDatabasePath + " (default)\n" +
+			want: "trees: database " + dbPath + " (default)\n" +
 				treesOpenLine + "\n" + treesOpenLine + "\n",
 		},
 		cmdInfo: {
 			run: func() error {
 				return runInfo(cancelledContext(t), io.Discard)
 			},
-			want: "info: database " + defaultDatabasePath + " (default)\n" +
+			want: "info: database " + dbPath + " (default)\n" +
 				infoOpenLine + "\n" + infoOpenLine + "\n",
 		},
 	}
@@ -716,6 +721,19 @@ func TestRunHelp(t *testing.T) {
 			t.Errorf("run(%v) stdout = %q, want nothing (data only)",
 				args, got)
 		}
+
+		// README §Location: the help of sfdupes and of scan says where
+		// the database is, and that XDG_STATE_HOME must be an absolute
+		// path.
+		for _, want := range []string{
+			"scan and dwim store", databaseEnv, rootDatabasePath,
+			stateHomeEnv, "~/.local/state", "not an absolute path is an error",
+		} {
+			if !strings.Contains(stderr.String(), want) {
+				t.Errorf("run(%v) stderr = %q, want it to name %q",
+					args, stderr.String(), want)
+			}
+		}
 	}
 
 	// The scan help lists --exclude with its argument.
@@ -758,28 +776,24 @@ func TestRunScanUsageLine(t *testing.T) {
 	}
 }
 
-func TestRunHelpDescribesDatabaseAndOutput(t *testing.T) {
+func TestRunHelpDescribesOutput(t *testing.T) {
 	t.Parallel()
 
-	// README §Settings: the help names the database, and says what
+	// README §Settings: the help of report, trees, info and dwim says
-	// report, trees, info and dwim print. The help of report and trees
+	// what each prints. The help of report and trees also explains
-	// also explains their PATH operands.
+	// their PATH operands. TestRunHelp checks the database paragraph
-	database := []string{
+	// of the root and scan help.
-		"scan and dwim store", databaseEnv, defaultDatabasePath, "needs root",
-	}
 	escapes := []string{`\\`, `\t`, `\n`, `\r`}
 	operands := []string{
 		"With PATH operands", "at or below them", "working directory",
 		"No PATH is looked up", "not followed",
 	}
 
-	// The subcommand whose help is printed; none for the root.
+	// The subcommand whose help is printed.
 	cases := []struct {
 		subcommand []string
 		want       []string
 	}{
-		{nil, database},
-		{[]string{cmdScan}, database},
 		{
 			[]string{cmdReport},
 			slices.Concat([]string{
@@ -999,8 +1013,7 @@ func TestRunScanSucceedsDespiteWarnings(t *testing.T) {
 
 	stderr := scanOK(t, dir)
 
-	warning := "hash " + unreadable + ": open " + unreadable +
+	warning := "scan: skipped " + unreadable + " (permission denied)\n"
-		": permission denied\n"
 	if !strings.Contains(stderr, warning) {
 		t.Errorf("stderr = %q, want %q", stderr, warning)
 	}
@@ -1272,7 +1285,7 @@ func TestRunScanSkipsSymlinkToFIFOOperand(t *testing.T) {
 
 	stderr := scanOK(t, link)
 
-	warning := "walk " + fifo + ": skipping FIFO operand\n"
+	warning := "scan: skipped " + fifo + " (a FIFO)\n"
 	if !strings.Contains(stderr, warning) {
 		t.Errorf("stderr = %q, want %q", stderr, warning)
 	}
@@ -1307,7 +1320,7 @@ func assertOperandSkipped(t *testing.T, dbPath, operand, kind,
 
 	stderr := scanOK(t, operand)
 
-	warning := "walk " + operand + ": skipping " + kind + " operand\n"
+	warning := "scan: skipped " + operand + " (a " + kind + ")\n"
 	if !strings.Contains(stderr, warning) {
 		t.Errorf("stderr = %q, want %q", stderr, warning)
 	}
@@ -1341,7 +1354,7 @@ func TestRunScanSkipsOperandInsideZFS(t *testing.T) {
 	writeFile(t, hourly, "f", pattern(1, 10))
 	f := writeFile(t, dir, "f", pattern(2, 10))
 
-	warning := "walk " + hourly + ": skipping operand inside a .zfs directory\n"
+	warning := "scan: skipped " + hourly + " (inside a .zfs directory)\n"
 
 	for _, tc := range []struct {
 		operands []string
@@ -1393,7 +1406,7 @@ func TestRunScanSkipsOperandThroughSymlinkIntoZFS(t *testing.T) {
 	stderr := scanOK(t, filepath.Join(snaps, "daily"))
 
 	daily := filepath.Join(dir, "pool", ".zfs", "snapshot", "daily")
-	warning := "walk " + daily + ": skipping operand inside a .zfs directory\n"
+	warning := "scan: skipped " + daily + " (inside a .zfs directory)\n"
 	summary := "scan: 0 files seen (0 added, 0 updated, 0 removed, " +
 		"0 unchanged), 1 skipped\n"
 
@@ -1612,6 +1625,100 @@ func TestRunDatabasePathUsedAsGiven(t *testing.T) {
 	})
 }
 
+func TestRunNonRootUsesStateDirectory(t *testing.T) {
+	// README §Location: without SFDUPES_DATABASE, a user other than
+	// root keeps the database in the XDG state directory. scan creates
+	// the missing directories with mode 0700, and report and trees read
+	// the same file.
+	if os.Geteuid() == 0 {
+		t.Skip("root uses " + rootDatabasePath)
+	}
+
+	t.Cleanup(xdg.Reload)
+
+	state := filepath.Join(t.TempDir(), "state")
+	t.Setenv(stateHomeEnv, state)
+	t.Setenv(databaseEnv, "")
+	xdg.Reload()
+
+	assertRunsUseDatabase(t, filepath.Join(state, "sfdupes"), "db.sqlite")
+
+	for _, dir := range []string{state, filepath.Join(state, "sfdupes")} {
+		fi, err := os.Stat(dir)
+		if err != nil {
+			t.Fatal(err)
+		}
+
+		if got := fi.Mode().Perm(); got != stateDirPerm {
+			t.Errorf("%s mode = %#o, want %#o", dir, got, stateDirPerm)
+		}
+	}
+}
+
+func TestRunNonRootDatabaseEnvDirectoryMode(t *testing.T) {
+	// README §Location: for SFDUPES_DATABASE, scan creates each missing
+	// directory with mode 755, for a user other than root too.
+	if os.Geteuid() == 0 {
+		t.Skip("needs a user other than root")
+	}
+
+	top := filepath.Join(t.TempDir(), "missing")
+	sub := filepath.Join(top, "sub")
+	t.Setenv(databaseEnv, filepath.Join(sub, "db.sqlite"))
+
+	scanOK(t, t.TempDir())
+
+	for _, dir := range []string{top, sub} {
+		fi, err := os.Stat(dir)
+		if err != nil {
+			t.Fatal(err)
+		}
+
+		if got := fi.Mode().Perm(); got != dbDirPerm {
+			t.Errorf("%s mode = %#o, want %#o", dir, got, dbDirPerm)
+		}
+	}
+}
+
+func TestRunRelativeStateHomeFails(t *testing.T) {
+	// README §Location: without SFDUPES_DATABASE, a user other than
+	// root whose XDG_STATE_HOME is not an absolute path gets an error
+	// naming it from every subcommand, which exits 1 before it does
+	// anything else.
+	if os.Geteuid() == 0 {
+		t.Skip("root does not use " + stateHomeEnv)
+	}
+
+	t.Setenv(stateHomeEnv, "relative/state")
+	t.Setenv(databaseEnv, "")
+
+	want := "sfdupes: " + stateHomeEnv +
+		" must be an absolute path, got \"relative/state\"\n"
+
+	// A scan that got past the check would stop at this missing
+	// operand, before it creates anything.
+	missing := filepath.Join(t.TempDir(), "missing")
+
+	for _, args := range [][]string{
+		{cmdScan, missing}, {cmdReport}, {cmdTrees}, {cmdInfo},
+	} {
+		var stdout, stderr bytes.Buffer
+
+		code := run(args, &stdout, &stderr)
+		if code != exitFatal || stderr.String() != want {
+			t.Errorf("run(%v) = %d with stderr %q, want %d and %q",
+				args, code, stderr.String(), exitFatal, want)
+		}
+	}
+
+	// dwim scans /, so runDwim gets the missing operand instead.
+	err := runDwim(t.Context(), missing, scanOptions{workers: 1}, io.Discard)
+	if !errors.Is(err, errStateHomeNotAbsolute) {
+		t.Errorf("runDwim: error = %v, want %v", err,
+			errStateHomeNotAbsolute)
+	}
+}
+
 // holdScanLock takes the lock on the database at path, as a running
 // scan does, and holds it until the test ends. It fails the test when
 // the lock is already held.

internal/sfdupes/db.go

@@ -15,28 +15,40 @@ import (
 	"strings"
 	"time"
 
+	"github.com/adrg/xdg"
 	"golang.org/x/sys/unix"
 	// The pure-Go SQLite driver, registered as "sqlite"; keeps cgo
 	// disabled.
 	_ "modernc.org/sqlite"
 )
 
-// defaultDatabasePath is where the persistent scan database lives when
+// rootDatabasePath is the database root uses when SFDUPES_DATABASE is
-// SFDUPES_DATABASE is not set.
+// not set.
-const defaultDatabasePath = "/var/lib/sfdupes/db.sqlite"
+const rootDatabasePath = "/var/lib/sfdupes/db.sqlite"
 
 // databaseEnv is the environment variable that overrides the database
 // path.
 const databaseEnv = "SFDUPES_DATABASE"
 
+// stateHomeEnv is the environment variable that names the XDG state
+// directory.
+const stateHomeEnv = "XDG_STATE_HOME"
+
 // schemaVersion is the database schema version this build reads and
 // writes, stored in PRAGMA user_version.
 const schemaVersion = 1
 
-// dbDirPerm is the mode for a database parent directory created by
+// dbDirPerm is the mode for a directory scan creates for root's
-// scan.
+// default database or for SFDUPES_DATABASE.
 const dbDirPerm = 0o755
 
+// stateDirPerm is the mode for a directory scan creates for the
+// database at a user's default path in the XDG state directory. It is
+// the mode the XDG Base Directory specification asks for, and it keeps
+// other users from reading the database, which lists every scanned
+// path.
+const stateDirPerm = 0o700
+
 // dbFilePerm is the mode for a database file created by scan, the mode
 // SQLite gives a database file it creates.
 const dbFilePerm = 0o644
@@ -144,15 +156,48 @@ var errScanRunning = errors.New("another scan is running")
 // 32 bytes of a SHA-256.
 var errHashSize = errors.New("hash is not 32 bytes")
 
+// errStateHomeNotAbsolute reports an XDG_STATE_HOME that is set to a
+// path that is not absolute.
+var errStateHomeNotAbsolute = errors.New(
+	stateHomeEnv + " must be an absolute path")
+
+// defaultDatabasePath is the database used when SFDUPES_DATABASE is not
+// set: rootDatabasePath for root (euid 0), and db.sqlite in the
+// directory sfdupes under the XDG state directory for any other user.
+// xdg reads XDG_STATE_HOME and HOME once, when the program starts. For
+// any other user, an XDG_STATE_HOME that is set and not an absolute
+// path is an error, where xdg would use its default instead. The path
+// is joined as text, not cleaned, for the reason absolutePath gives.
+func defaultDatabasePath(euid int) (string, error) {
+	if euid == 0 {
+		return rootDatabasePath, nil
+	}
+
+	stateHome := os.Getenv(stateHomeEnv)
+	if stateHome != "" && !filepath.IsAbs(stateHome) {
+		return "", fmt.Errorf("%w, got %q", errStateHomeNotAbsolute,
+			stateHome)
+	}
+
+	stateDir := strings.TrimSuffix(xdg.StateHome, "/")
+
+	return stateDir + "/sfdupes/db.sqlite", nil
+}
+
 // databasePath resolves the database location: SFDUPES_DATABASE when
-// set and non-empty, the compiled-in default otherwise. It returns the
+// set and non-empty, defaultDatabasePath otherwise. It returns the
 // path, made absolute by absolutePath, and where the path came from,
 // "default" or "from SFDUPES_DATABASE", for the line each subcommand
 // prints first on stderr.
 func databasePath() (string, string, error) {
-	path, source := defaultDatabasePath, "default"
+	path, source := os.Getenv(databaseEnv), "from "+databaseEnv
-	if p := os.Getenv(databaseEnv); p != "" {
+	if path == "" {
-		path, source = p, "from "+databaseEnv
+		def, err := defaultDatabasePath(os.Geteuid())
+		if err != nil {
+			return "", "", err
+		}
+
+		path, source = def, "default"
 	}
 
 	abs, err := absolutePath(path)
@@ -188,15 +233,23 @@ func absolutePath(path string) (string, error) {
 
 // createDatabaseDirectory creates the directory the database at path
 // goes in, and any directory above it, if missing. path is absolute, as
-// databasePath returns it. The directory is path with its last element
+// databasePath returns it. Each directory gets mode dbDirPerm, or
-// cut off as text, by filepath.Split. filepath.Dir would clean it,
+// stateDirPerm when databasePath uses a user's default path in the XDG
-// dropping each ".." together with the directory name before it; when
+// state directory: the user is not root and SFDUPES_DATABASE is unset
-// that name is a symlink, the cleaned path names another directory than
+// or empty. The directory is path with its last element cut off as
-// the one the system reaches when it opens path.
+// text, by filepath.Split. filepath.Dir would clean it, dropping each
+// ".." together with the directory name before it; when that name is a
+// symlink, the cleaned path names another directory than the one the
+// system reaches when it opens path.
 func createDatabaseDirectory(path string) error {
 	dir, _ := filepath.Split(path)
 
-	err := os.MkdirAll(dir, dbDirPerm)
+	perm := os.FileMode(dbDirPerm)
+	if os.Geteuid() != 0 && os.Getenv(databaseEnv) == "" {
+		perm = stateDirPerm
+	}
+
+	err := os.MkdirAll(dir, perm)
 	if err != nil {
 		return fmt.Errorf("create database directory: %w", err)
 	}

internal/sfdupes/db_test.go

@@ -9,10 +9,13 @@ import (
 	"io/fs"
 	"os"
 	"path/filepath"
+	"runtime"
 	"slices"
 	"strings"
 	"testing"
 	"time"
+
+	"github.com/adrg/xdg"
 )
 
 // testDBPath returns a database path inside a fresh temp dir, with
@@ -56,11 +59,16 @@ func TestDatabasePath(t *testing.T) {
 
 	t.Chdir(dir)
 
+	defaultPath, err := defaultDatabasePath(os.Geteuid())
+	if err != nil {
+		t.Fatal(err)
+	}
+
 	fromEnv := "from " + databaseEnv
 	cases := []struct {
 		env, path, source string
 	}{
-		{"", defaultDatabasePath, "default"},
+		{"", defaultPath, "default"},
 		{"/custom/place.sqlite", "/custom/place.sqlite", fromEnv},
 		{"/custom/link/../place.sqlite", "/custom/link/../place.sqlite",
 			fromEnv},
@@ -89,6 +97,48 @@ func TestDatabasePath(t *testing.T) {
 	}
 }
 
+func TestDefaultDatabasePath(t *testing.T) {
+	// README §Location: root's database is /var/lib/sfdupes/db.sqlite.
+	// Any other user's is db.sqlite in sfdupes under XDG_STATE_HOME, or
+	// under ~/.local/state when XDG_STATE_HOME is unset or empty. For
+	// that user, an XDG_STATE_HOME that is not an absolute path is an
+	// error.
+	if runtime.GOOS != goosLinux {
+		t.Skip("macOS has another default state directory")
+	}
+
+	t.Cleanup(xdg.Reload)
+	t.Setenv("HOME", "/home/someone")
+
+	cases := []struct {
+		euid      int
+		stateHome string
+		want      string
+		wantErr   error
+	}{
+		{0, "", rootDatabasePath, nil},
+		{0, "/elsewhere/state", rootDatabasePath, nil},
+		{0, "relative/state", rootDatabasePath, nil},
+		{1000, "", "/home/someone/.local/state/sfdupes/db.sqlite", nil},
+		{1000, "relative/state", "", errStateHomeNotAbsolute},
+		{1000, "~/state", "", errStateHomeNotAbsolute},
+		{1000, "/elsewhere/state", "/elsewhere/state/sfdupes/db.sqlite", nil},
+		{1000, "/elsewhere/state/", "/elsewhere/state/sfdupes/db.sqlite", nil},
+	}
+
+	for _, tc := range cases {
+		t.Setenv(stateHomeEnv, tc.stateHome)
+		xdg.Reload()
+
+		got, err := defaultDatabasePath(tc.euid)
+		if got != tc.want || !errors.Is(err, tc.wantErr) {
+			t.Errorf("defaultDatabasePath(%d) with %s=%q = %q, %v; want %q, %v",
+				tc.euid, stateHomeEnv, tc.stateHome, got, err, tc.want,
+				tc.wantErr)
+		}
+	}
+}
+
 func TestOpenScanDatabaseCreates(t *testing.T) {
 	t.Parallel()
 

internal/sfdupes/dwim_test.go

@@ -121,9 +121,9 @@ func TestRunDwimExcludeLeavesTreeOut(t *testing.T) {
 		t.Errorf("group = %q, want it to start %q", lines[1], want)
 	}
 
-	skipped := "scan: skipped " + excluded + " (--exclude)\n"
+	leftOut := "scan: left out " + excluded + " (--exclude)\n"
-	if got := stderr(); !strings.Contains(got, skipped) {
+	if got := stderr(); !strings.Contains(got, leftOut) {
-		t.Errorf("stderr = %q, want it to contain %q", got, skipped)
+		t.Errorf("stderr = %q, want it to contain %q", got, leftOut)
 	}
 }
 

internal/sfdupes/exclude.go

@@ -72,14 +72,31 @@ type walkRules struct {
 func defaultExcludes(goos string) []string {
 	switch goos {
 	case goosLinux:
-		return []string{"/proc", "/sys", "/dev", "/run", "/tmp", "/var/tmp"}
+		// The kernel's view of the running system, temporary files, and
+		// caches that programs and package managers fill again on their
+		// own: the system's, root's and those in each home in /home.
+		return []string{
+			"/proc", "/sys", "/dev", "/run", "/tmp", "/var/tmp", "/var/cache",
+			"/root/.cache", "/root/.npm", "/root/.cargo/registry",
+			"/root/go/pkg/mod",
+			"/home/*/.cache", "/home/*/.npm", "/home/*/.cargo/registry",
+			"/home/*/go/pkg/mod",
+		}
 	case goosDarwin:
 		// A walk of / reaches everything in /System/Volumes/Data through
 		// the firmlinks, so walking it as well would record each of
-		// those files twice.
+		// those files twice. /private/var/folders holds each user's
+		// temporary files and caches. The walk reaches root's home,
+		// /var/root, as /private/var/root.
 		return []string{
 			"/dev", "/private/tmp", "/private/var/tmp", "/private/var/vm",
 			"/System/Volumes/VM", "/System/Volumes/Data",
+			"/private/var/folders", "/Library/Caches",
+			"/private/var/root/Library/Caches", "/private/var/root/.cache",
+			"/private/var/root/.npm", "/private/var/root/.cargo/registry",
+			"/private/var/root/go/pkg/mod",
+			"/Users/*/Library/Caches", "/Users/*/.cache", "/Users/*/.npm",
+			"/Users/*/.cargo/registry", "/Users/*/go/pkg/mod",
 		}
 	case goosWindows:
 		// Relative to the root of the drive an operand lies on. The
@@ -186,7 +203,7 @@ func (rules walkRules) excludePathLine(path string, fi fs.FileInfo) string {
 		return ""
 	}
 
-	return fmt.Sprintf("scan: skipped %s (--exclude)", path)
+	return fmt.Sprintf("scan: left out %s (--exclude)", path)
 }
 
 // nonPersistentFilesystemTypes returns the filesystem types that are
@@ -216,7 +233,7 @@ func (rules walkRules) nonPersistentLine(path string, fi fs.FileInfo) string {
 		return ""
 	}
 
-	return fmt.Sprintf("scan: skipped %s (%s, not a persistent filesystem)",
+	return fmt.Sprintf("scan: left out %s (%s, not a persistent filesystem)",
 		path, fsType)
 }
 

internal/sfdupes/exclude_test.go

@@ -24,10 +24,22 @@ func TestDefaultExcludes(t *testing.T) {
 	// on any other system. The Windows list exists in the code only: the
 	// binary does not build for Windows, so the README leaves it out.
 	cases := map[string][]string{
-		goosLinux: {"/proc", "/sys", "/dev", "/run", "/tmp", "/var/tmp"},
+		goosLinux: {
+			"/proc", "/sys", "/dev", "/run", "/tmp", "/var/tmp", "/var/cache",
+			"/root/.cache", "/root/.npm", "/root/.cargo/registry",
+			"/root/go/pkg/mod",
+			"/home/*/.cache", "/home/*/.npm", "/home/*/.cargo/registry",
+			"/home/*/go/pkg/mod",
+		},
 		goosDarwin: {
 			"/dev", "/private/tmp", "/private/var/tmp", "/private/var/vm",
 			"/System/Volumes/VM", "/System/Volumes/Data",
+			"/private/var/folders", "/Library/Caches",
+			"/private/var/root/Library/Caches", "/private/var/root/.cache",
+			"/private/var/root/.npm", "/private/var/root/.cargo/registry",
+			"/private/var/root/go/pkg/mod",
+			"/Users/*/Library/Caches", "/Users/*/.cache", "/Users/*/.npm",
+			"/Users/*/.cargo/registry", "/Users/*/go/pkg/mod",
 		},
 		goosWindows: {
 			"pagefile.sys", "swapfile.sys", "hiberfil.sys",
@@ -63,6 +75,29 @@ func TestIsDefaultExclude(t *testing.T) {
 			t.Errorf("isDefaultExclude(%q) = %v, want %v", path, got, want)
 		}
 	}
+
+	// The Linux list matches the cache directories in /root and in each
+	// home in /home, but nothing below them and no .cache directory
+	// elsewhere.
+	linux := map[string]bool{
+		"/home/u/.cache":           true,
+		"/home/u/go/pkg/mod":       true,
+		"/root/.cache":             true,
+		"/var/cache":               true,
+		"/home/u/.cache/pip":       false,
+		"/home/u/src/.cache":       false,
+		"/srv/home/u/.cache":       false,
+		"/home/u/.cargo/bin":       false,
+		"/home/u/go/pkg/mod/cache": false,
+	}
+
+	for path, want := range linux {
+		got := isDefaultExclude(path, defaultExcludes(goosLinux))
+		if got != want {
+			t.Errorf("isDefaultExclude(%q, Linux list) = %v, want %v", path,
+				got, want)
+		}
+	}
 }
 
 func TestReadMountInfo(t *testing.T) {
@@ -207,7 +242,7 @@ func TestScanDoesNotEnterDefaultExclude(t *testing.T) {
 		t.Errorf("%s recorded: the walk entered %s", unseen, excluded)
 	}
 
-	line := "\nscan: skipped " + excluded + " (default exclude)\n"
+	line := "\nscan: left out " + excluded + " (default exclude)\n"
 	if !strings.Contains(stderr(), line) {
 		t.Errorf("stderr = %q, want %q", stderr(), line)
 	}
@@ -276,7 +311,7 @@ func TestScanNamesNoDefaultExcludeThatIsAnOperand(t *testing.T) {
 
 	recordByPath(t, dbRecords(t, db), f)
 
-	if strings.Contains(stderr(), "scan: skipped") {
+	if strings.Contains(stderr(), "scan: left out") {
 		t.Errorf("stderr = %q, want no line for %s", stderr(), excluded)
 	}
 }
@@ -290,15 +325,15 @@ func TestRunScanTempDirOperand(t *testing.T) {
 	t.Setenv(databaseEnv, path)
 
 	dir := realTempDir(t)
-	if !underAnyRoot(dir, defaultExcludes(runtime.GOOS)) {
+	if !insideDefaultExclude(dir, defaultExcludes(runtime.GOOS)) {
 		t.Skip(dir + " is not inside a default exclude")
 	}
 
 	f := writeFile(t, dir, "sub/f", pattern(1, 10))
 
 	stderr := scanOK(t, dir)
-	if strings.Contains(stderr, "scan: skipped") {
+	if strings.Contains(stderr, "scan: left out") {
-		t.Errorf("stderr = %q, want no line saying a path was skipped",
+		t.Errorf("stderr = %q, want no line saying a path was left out",
 			stderr)
 	}
 
@@ -342,7 +377,7 @@ func TestRunScanNoDefaultExcludes(t *testing.T) {
 	path := testDBPath(t)
 	t.Setenv(databaseEnv, path)
 
-	line := "scan: skipped /var/tmp (default exclude)\n"
+	line := "scan: left out /var/tmp (default exclude)\n"
 
 	if stderr := scanOK(t, "/var"); !strings.Contains(stderr, line) {
 		t.Errorf("scan /var: stderr = %q, want %q", stderr, line)
@@ -383,7 +418,7 @@ func TestRunScanExcludeNested(t *testing.T) {
 
 	stderr := scanOK(t, excludeFlag, excluded, dir)
 
-	line := "scan: skipped " + excluded + " (--exclude)\n"
+	line := "scan: left out " + excluded + " (--exclude)\n"
 	if !strings.Contains(stderr, line) {
 		t.Errorf("stderr = %q, want %q", stderr, line)
 	}
@@ -445,8 +480,8 @@ func TestRunScanExcludeReachedByAnotherPath(t *testing.T) {
 	stderr := scanOK(t, excludeFlag, filepath.Join(link, "cache"),
 		excludeFlag, dotDot, dir)
 
-	for _, skipped := range []string{filepath.Dir(inDir), file, hardLink} {
+	for _, leftOut := range []string{filepath.Dir(inDir), file, hardLink} {
-		line := "scan: skipped " + skipped + " (--exclude)\n"
+		line := "scan: left out " + leftOut + " (--exclude)\n"
 		if !strings.Contains(stderr, line) {
 			t.Errorf("stderr = %q, want %q", stderr, line)
 		}
@@ -580,7 +615,7 @@ func TestRunScanSkipsNonPersistentOperand(t *testing.T) {
 
 	t.Setenv(databaseEnv, testDBPath(t))
 
-	line := "scan: skipped " + shm + " (" + fsType +
+	line := "scan: left out " + shm + " (" + fsType +
 		", not a persistent filesystem)\n"
 	summary := "scan: 0 files seen"
 
@@ -634,7 +669,7 @@ func TestScanKeepsRecordsOnNonPersistent(t *testing.T) {
 		t.Errorf("%s recorded", unseen)
 	}
 
-	line := "scan: skipped " + shm + " (" + fsType +
+	line := "scan: left out " + shm + " (" + fsType +
 		", not a persistent filesystem)\n"
 	if !strings.Contains(stderr(), line) {
 		t.Errorf("stderr = %q, want %q", stderr(), line)
@@ -667,7 +702,7 @@ func TestWalkLeavesOutNonPersistent(t *testing.T) {
 
 		leftOut = append(leftOut, ev.leftOut)
 
-		want := "scan: skipped " + ev.leftOut + " (" + fsType +
+		want := "scan: left out " + ev.leftOut + " (" + fsType +
 			", not a persistent filesystem)"
 		if ev.warn != want {
 			t.Errorf("line = %q, want %q", ev.warn, want)
@@ -711,6 +746,20 @@ func shmTempDir(t *testing.T) (string, string, walkRules) {
 	return dir, fsType, walkRules{nonPersistent: nonPersistent}
 }
 
+// insideDefaultExclude reports whether path, which is absolute and
+// clean, or a directory above it is one of excludes.
+func insideDefaultExclude(path string, excludes []string) bool {
+	for dir := path; ; dir = filepath.Dir(dir) {
+		if isDefaultExclude(dir, excludes) {
+			return true
+		}
+
+		if dir == "/" {
+			return false
+		}
+	}
+}
+
 // assertRecorded checks that the database at dbPath holds a record for
 // path when want is set, and none when it is not.
 func assertRecorded(t *testing.T, dbPath, path string, want bool) {

internal/sfdupes/scan.go

@@ -664,8 +664,8 @@ func isDatabaseFile(f fileRec, rules walkRules) bool {
 // inode run, committing completed records to the database in batches
 // as results arrive, so a long scan persists its progress as it goes
 // (an interrupted scan resumes cheaply: the next run skips everything
-// already recorded). A run that fails to hash is warned about and
+// already recorded). Each path of a run that fails to hash is warned
-// counted as skipped; stale records for its paths, if any, are deleted
+// about and counted as skipped; stale records for them, if any, are deleted
 // by the update phase. A run whose file no longer exists gets no record
 // and no warning, and is not counted as skipped. The total counts actual
 // reads, so the display's time remaining is meaningful.
@@ -706,9 +706,12 @@ func (s *scanState) hashPhase(ctx context.Context, workers int) error {
 		}
 
 		if r.err != nil {
-			s.st.skipped += len(r.run)
+			// Every path in the run is counted, so every path gets a line.
+			for _, rec := range r.run {
+				s.st.skipped++
 
-			prog.warnf("hash %s: %v", r.run[0].path, r.err)
+				prog.warnf("%s", skippedLine(rec.path, r.err))
+			}
 
 			continue
 		}
@@ -915,8 +918,7 @@ func operandWarning(root string, fi fs.FileInfo) string {
 	// The walk never enters a .zfs directory, so it starts from nothing
 	// inside one either.
 	if slices.Contains(strings.Split(filepath.Dir(root), "/"), ".zfs") {
-		return fmt.Sprintf("walk %s: skipping operand inside a .zfs directory",
+		return fmt.Sprintf("scan: skipped %s (inside a .zfs directory)", root)
-			root)
 	}
 
 	var kind string
@@ -942,7 +944,21 @@ func operandWarning(root string, fi fs.FileInfo) string {
 		kind = "non-regular file"
 	}
 
-	return fmt.Sprintf("walk %s: skipping %s operand", root, kind)
+	return fmt.Sprintf("scan: skipped %s (a %s)", root, kind)
+}
+
+// skippedLine returns the warning for a path the scan skips because of
+// err. The reason is the error without its path, which the line already
+// names.
+func skippedLine(path string, err error) string {
+	reason := err.Error()
+
+	var pathErr *fs.PathError
+	if errors.As(err, &pathErr) {
+		reason = pathErr.Err.Error()
+	}
+
+	return fmt.Sprintf("scan: skipped %s (%s)", path, reason)
 }
 
 // seedRoot turns one PATH operand into the walk's starting state: a
@@ -965,7 +981,7 @@ func seedRoot(ctx context.Context, root string, rules walkRules,
 
 	if err != nil {
 		sendEvent(ctx, events, walkEvent{
-			warn: fmt.Sprintf("walk %s: %v", root, err),
+			warn: skippedLine(root, err),
 			fail: true,
 		})
 
@@ -1094,7 +1110,7 @@ func walkOneDir(ctx context.Context, job dirJob, oneFS bool,
 
 	if err != nil {
 		sendEvent(ctx, events, walkEvent{
-			warn: fmt.Sprintf("walk %s: %v", job.path, err),
+			warn: skippedLine(job.path, err),
 			fail: true,
 		})
 
@@ -1153,7 +1169,7 @@ func emitFile(ctx context.Context, p string, e fs.DirEntry,
 
 	if err != nil {
 		sendEvent(ctx, events, walkEvent{
-			warn: fmt.Sprintf("stat %s: %v", p, err),
+			warn: skippedLine(p, err),
 			fail: true,
 		})
 
@@ -1214,7 +1230,7 @@ func subdirJob(ctx context.Context, p string, e fs.DirEntry,
 
 	if err != nil {
 		sendEvent(ctx, events, walkEvent{
-			warn: fmt.Sprintf("walk %s: %v", p, err),
+			warn: skippedLine(p, err),
 			fail: true,
 		})
 
@@ -1235,7 +1251,7 @@ func subdirJob(ctx context.Context, p string, e fs.DirEntry,
 	if isDefaultExclude(p, rules.defaultExcludes) {
 		sendEvent(ctx, events, walkEvent{
 			leftOut: p,
-			warn:    fmt.Sprintf("scan: skipped %s (default exclude)", p),
+			warn:    fmt.Sprintf("scan: left out %s (default exclude)", p),
 		})
 
 		return dirJob{}, false

internal/sfdupes/scan_test.go

@@ -752,8 +752,9 @@ func TestWalkPassesOverRemovedPaths(t *testing.T) {
 
 // TestScanWarnsAboutUnreadablePaths checks that a permission error is
 // still warned about and counted as skipped, unlike a removed path. The
-// walk reaches an unreadable directory, and a file in a directory that
+// walk reaches an unreadable directory. It also reaches a file and a
-// can be read but not searched. Another such file is an operand.
+// subdirectory in a directory that can be read but not searched. Another
+// such file is an operand.
 //
 //nolint:paralleltest // captureStderr replaces the process-wide os.Stderr
 func TestScanWarnsAboutUnreadablePaths(t *testing.T) {
@@ -765,6 +766,7 @@ func TestScanWarnsAboutUnreadablePaths(t *testing.T) {
 	dir := t.TempDir()
 	tree := filepath.Join(dir, "tree")
 	walked := writeFile(t, tree, "unsearchable/f", []byte("x"))
+	subdir := filepath.Join(filepath.Dir(walked), "d")
 	operand := writeFile(t, dir, "unsearchable/f", []byte("y"))
 	unreadable := filepath.Join(tree, "unreadable")
 
@@ -773,6 +775,11 @@ func TestScanWarnsAboutUnreadablePaths(t *testing.T) {
 		t.Fatal(err)
 	}
 
+	err = os.Mkdir(subdir, 0o750)
+	if err != nil {
+		t.Fatal(err)
+	}
+
 	modes := map[string]fs.FileMode{
 		unreadable:            0,
 		filepath.Dir(walked):  0o400,
@@ -793,15 +800,16 @@ func TestScanWarnsAboutUnreadablePaths(t *testing.T) {
 	}
 
 	st := syncTree(t, openTestDB(t), tree, operand)
-	if st != (scanStats{skipped: 3}) {
+	if st != (scanStats{skipped: 4}) {
-		t.Errorf("stats = %+v, want 3 skipped and nothing else", st)
+		t.Errorf("stats = %+v, want 4 skipped and nothing else", st)
 	}
 
 	got := stderr()
 	for _, warning := range []string{
-		"walk " + unreadable + ": open " + unreadable + ": permission denied\n",
+		"scan: skipped " + unreadable + " (permission denied)\n",
-		"stat " + walked + ": lstat " + walked + ": permission denied\n",
+		"scan: skipped " + walked + " (permission denied)\n",
-		"walk " + operand + ": lstat " + operand + ": permission denied\n",
+		"scan: skipped " + subdir + " (permission denied)\n",
+		"scan: skipped " + operand + " (permission denied)\n",
 	} {
 		if !strings.Contains(got, warning) {
 			t.Errorf("stderr = %q, want %q", got, warning)
@@ -1941,14 +1949,19 @@ func TestScanHardlinksReadOnce(t *testing.T) {
 	}
 }
 
+// TestScanHardlinkRunFailsTogether makes a file with two hard links
+// unreadable. Both paths are counted as skipped, and each gets its own
+// warning.
+//
+//nolint:paralleltest // captureStderr replaces the process-wide os.Stderr
 func TestScanHardlinkRunFailsTogether(t *testing.T) {
-	t.Parallel()
+	stderr := captureStderr(t)
-
 	dir := t.TempDir()
 	db := openTestDB(t)
 	a := writeFile(t, dir, "a.bin", pattern(1, 300))
+	b := filepath.Join(dir, "b.bin")
 
-	err := os.Link(a, filepath.Join(dir, "b.bin"))
+	err := os.Link(a, b)
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -1964,6 +1977,15 @@ func TestScanHardlinkRunFailsTogether(t *testing.T) {
 	if st.skipped != 2 || st.added != 0 {
 		t.Fatalf("stats = %+v, want both hardlink paths skipped", st)
 	}
+
+	got := stderr()
+
+	for _, p := range []string{a, b} {
+		warning := "scan: skipped " + p + " (permission denied)\n"
+		if !strings.Contains(got, warning) {
+			t.Errorf("stderr = %q, want %q", got, warning)
+		}
+	}
 }
 
 // TestHashPhaseSkipsFileReplacedByFIFO replaces a walked regular file
@@ -2001,7 +2023,7 @@ func TestHashPhaseSkipsFileReplacedByFIFO(t *testing.T) {
 			s.st.skipped, len(s.batch))
 	}
 
-	warning := "hash " + p + ": not a regular file\n"
+	warning := "scan: skipped " + p + " (not a regular file)\n"
 	if got := stderr(); strings.Count(got, warning) != 1 {
 		t.Errorf("stderr = %q, want %q once", got, warning)
 	}